Rethinking Efficient Mixture-of-Experts for Remote Sensing Modality-Missing Classification

TL;DR

This paper proposes MaMOL, a novel parameter-efficient Mixture-of-Experts framework that dynamically adapts to missing modalities in remote sensing classification, improving robustness and generalization across various scenarios.

Contribution

It introduces a dual-routing MoE model that unifies multiple missing-modality cases, enhancing adaptability and efficiency in remote sensing tasks.

Findings

MaMOL outperforms existing methods in missing-modality scenarios.

It achieves significant robustness improvements with minimal computational overhead.

Transfer experiments show strong scalability and cross-domain applicability.

Abstract

Multimodal remote sensing classification often suffers from missing modalities caused by sensor failures and environmental interference, leading to severe performance degradation. In this work, we rethink missing-modality learning from a conditional computation perspective and investigate whether Mixture-of-Experts (MoE) models can inherently adapt to diverse modality-missing scenarios. We first conduct a systematic study of representative MoE paradigms under various missing-modality settings, revealing both their potential and limitations. Building on these insights, we propose a Missing-aware Mixture-of-LoRAs (MaMOL), a parameter-efficient MoE framework that unifies multiple modality-missing cases within a single model. MaMOL introduces a dual-routing mechanism to decouple modality-invariant shared experts and modality-aware dynamic experts, enabling automatic expert activation conditioned on available modalities. Extensive experiments on multiple remote sensing benchmarks demonstrate that MaMOL significantly improves robustness and generalization under diverse missing-modality scenarios with minimal computational overhead. Transfer experiments on natural image datasets further validate its scalability and cross-domain applicability.